An Explainable Cryobiopsy Artificial Intelligence (AI) Model, CRAIO, to Predict Progression in Interstitial Pneumonia

Interstitial lung disease encompasses diverse pulmonary disorders with varied prognoses. Current pathological diagnoses suffer from interobserver variability, necessitating more standardized approaches. We developed an ensemble cryobiopsy artificial intelligence model (CRAIO) to analyze transbronchial lung cryobiopsy specimens and predict patient outcomes. CRAIO comprises 7 modules for detecting histological features, generating 17 pathologically significant findings. A downstream XGBoost classifier was developed to predict disease progression using these findings. The model's performance was evaluated using respiratory function changes and survival analysis in cross-validation and external test cohorts. In the internal cross-validation (135 cases), the model predicted 105 cases without disease progression and 30 cases with disease progression. The annual absolute change in % forced vital capacity was -1.293 in the nonprogressive group versus -5.198 in the progressive group, a difference that was significant for CRAIO, while only 5 of 19 pathologists achieved significant differentiation using usual interstitial pneumonia diagnosis. Survival analysis demonstrated significantly shorter survival times in the progressive group (P = .038). CRAIO provides a comprehensive, interpretable approach to analyzing transbronchial lung cryobiopsy specimens, offering potential for standardizing interstitial lung disease diagnosis and predicting disease progression. The model could facilitate early identification of progressive cases and guide personalized therapeutic interventions.